Comparison of three rating scales as outcome measures for treatment trials of depression in Alzheimer disease: Findings from DIADS

Lawrence S. Mayer, R. Curt Bay, Antonios Politis, Martin Steinberg, Cynthia Steele, Alva S. Baker, Peter V. Rabins, Constantine G. Lyketsos

Research output: Contribution to journalArticlepeer-review

34 Scopus citations


Context: Major depression affects about 25% of patients with Alzheimer's disease (AD) and has serious adverse consequences for patients as well as caregivers. Studies of treatments for depression in AD, like most treatment studies, depend on the ability of the scales used to measure outcome to detect a difference between the effects of treatment and control, particularly in trials conducted over waves. Objective: To compare the ability of three depression scales, and some of their subscales, to detect the difference in the effects of drug (treatment) and placebo (control). Design: Comparison of three scales of depression in terms of percent variance explained as indicated by the adjusted or partial eta-squared for the effect of drug versus placebo, controlling for baseline depression, in a randomized, placebo-controlled, parallel, 12-week, clinical trial of sertraline for the treatment of depression with AD. Setting: University outpatient clinic. Participants: Forty-four patients with probable Alzheimer's disease and Major Depressive Episode. Outcome measures: The Cornell Scale for Depression in Dementia (CSDD), the Hamilton Depression Rating Scale (HDRS), and the Neuropsychiatric-Inventory Mood Domains (NPI-M). Results: Examination of the treatment effects as indicated by the partial eta-squared's for each scale at each wave, revealed a slight, but not significant, advantage for the use of the CSDD over the HDRS, and a significant advantage for the use of either of these over the NPI-M. Treatment effects, as reflected in the partial eta-squared's computed for the subscales at each wave, were significant for all four subscales, and were largest for the CSDD 'mood' subscale although they were not significantly greater than for the other subscales. Conclusions: The CSDD, and particularly its mood subscale, appears to be more sensitive than the HDRS, it's subscales or the NPI-M, for comparing drug to placebo in treating major depression in AD patients. Treatment effects as reflected in the partial eta-squared's were largest on the CSDD mood subscale and increased over time. The pattern for the other subscales was non-monotonic over waves and resembled the pattern for the entire scale. Perhaps combining the CSDD two subscales obscures the treatment effects for the separate subscales.

Original languageEnglish (US)
Pages (from-to)930-936
Number of pages7
JournalInternational journal of geriatric psychiatry
Issue number10
StatePublished - Oct 2006


  • Alzheimer's disease
  • Clinical trials
  • Cornell Scale for Depression in Dementia
  • Depression in Alzheimer's disease
  • Effect size
  • Hamilton Depression Rating Scale
  • Neuropsychiatry-Inventory
  • Partial eta-squared

ASJC Scopus subject areas

  • Geriatrics and Gerontology
  • Psychiatry and Mental health


Dive into the research topics of 'Comparison of three rating scales as outcome measures for treatment trials of depression in Alzheimer disease: Findings from DIADS'. Together they form a unique fingerprint.

Cite this